P.N. Seneviratne
J.F. Morrall

Short term manual counts are used in many pedestrian studies to overcome time, cost, and equipment constraints. Depending on the count period, several techniques are available for expanding short-term counts with the objective of estimating pedestrian volume over extended periods of time. When flow is highly variable and only a few short-term counts can be conducted, the expanded values can contain large margins of error. This paper presents the findings of two studies on the development and verification of models for expanding short-term (five-, 10-, 15-minute) counts. Using data from the Central Business Districts of Calgary and Montreal, it is shown that such models can explain up to 90% of the variation in expected volumes. The paper also presents a methodology based on Bayesian theory for minimizing errors resulting from short-term counts with large variances. The methodology, which takes into consideration the variance as well as the number of short-term counts, also enables planners to update the short-term counts on the basis of subsequent surveys.